# Centos8下 K8S 安装配置Mysql8 多主MGR > 作者/来源: UCloud 运营管理员 > 发布时间: 2023-01-11T05:20:00.000Z > 分类: 其他 > 标签: AI, Docker > 原文链接: http://117.50.162.249:3000/yun/articles/888 --- # Centos8下 K8S 安装配置Mysql8 多主MGR > 来源: https://www.ucloud.cn/yun/130047.html > 作者: IT那活儿 > 发布日期: 发布于2023-01-11 13:20 Centos8下 K8S 安装配置Mysql8 多主MGR ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_000.png) 一.环境配置 k8s环境：1台master节点，3台node节点。 | | | --- | | 192.168.37.128 k8s1 192.168.37.130 k8s2 192.168.37.131 k8s3 192.168.37.129 k8s4 ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_001.png) | 二.搭建步骤 ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_011.png) 1、准备MySQL镜像 | | | --- | | 拉取MySQL最新的镜像 docker pull mysql | ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_011.png) 2、准备相关yaml文件 **准备namespace文件** | | | --- | | cat namespace.yaml apiVersion: v1 kind: Namespace metadata:  name: mysqldb | **准备service文件** | | | --- | | cat service.yaml apiVersion: v1 kind: Service metadata: name: mgrtest  namespace: mysqldb spec: clusterIP: None  selector:   name: mysql-mgr   ports:  - name: foo    port: 3306    targetPort: 3306 | **MySQL参数配置文件configmap(节点1)** | | | --- | | 节点1： cat mysql-mgr-cnf-0.yaml apiVersion: v1 kind: ConfigMap metadata:  name: mysql-mgr-0-cnf namespace: mysqldb data:  mysql-mgr-0.cnf: |   [mysqld]    port = 3306    character\_set\_server = utf8    server\_id = 1    log\_bin = mysql-bin   relay-log = relay-bin    #back\_log = 500   #max\_connections = 3000    #wait\_timeout = 5022397   interactive\_timeout = 5022397    max\_connect\_errors = 1000   relay-log-recovery=1    #max\_allowed\_packet = 32M   sort\_buffer\_size = 4M    read\_buffer\_size = 4M   join\_buffer\_size = 8M    thread\_cache\_size = 64   #tmp\_table\_size = 256M    log\_slave\_updates=1   long\_query\_time = 1    slow\_query\_log = 1   skip-name-resolve    innodb\_buffer\_pool\_size=700M   #innodb\_data\_file\_path = ibdata1:1024M:autoextend   innodb\_flush\_log\_at\_trx\_commit=1    innodb\_log\_buffer\_size = 16M    innodb\_log\_file\_size = 256M   innodb\_log\_files\_in\_group = 2    innodb\_max\_dirty\_pages\_pct = 50    sync\_binlog=1    master\_info\_repository=TABLE   relay\_log\_info\_repository=TABLE    log\_timestamps=SYSTEM   gtid\_mode = ON    enforce\_gtid\_consistency = ON   master\_info\_repository = TABLE    relay\_log\_info\_repository = TABLE    log\_slave\_updates = ON    binlog\_checksum = NONE    log\_slave\_updates = ON   slave\_parallel\_type=LOGICAL\_CLOCK   slave\_parallel\_workers=8    slave-preserve-commit-order=on   #group\_replication\_compression\_threshold=200000   plugin\_load\_add=group\_replication.so   transaction\_write\_set\_extraction = XXHASH64   loose-group\_replication\_group\_name="01e5fb97-be64-41f7-bafd-3afc7a6ab555"   loose-group\_replication\_start\_on\_boot=off   loose-group\_replication\_local\_address="k8s2.mgrtest.mysqldb.svc.cluster.local:13306"   loose-group\_replication\_group\_seeds="k8s2.mgrtest.mysqldb.svc.cluster.local:13306,k8s3.mgrtest.mysqldb.svc.cluster.local:13306,k8s4.mgrtest.mysqldb.svc.cluster.local:13306"   loose-group\_replication\_bootstrap\_group = off   loose-group\_replication\_ip\_whitelist=10.96.0.0/16,172.16.0.0/16,192.168.0.0/16,172.17.0.0/16,127.0.0.1/8   report\_host = k8s2.mgrtest.mysqldb.svc.cluster.local   [mysqldump]    quick    max\_allowed\_packet = 32M | **MySQL参数配置文件configmap(节点2)** | | | --- | | cat mysql-mgr-cnf-1.yaml   apiVersion: v1 kind: ConfigMap metadata:  name: mysql-mgr-1-cnf namespace: mysqldb data:  mysql-mgr-1.cnf: |   [mysqld]    port = 3306    character\_set\_server = utf8    server\_id = 2    log\_bin = mysql-bin   relay-log = relay-bin    #back\_log = 500   #max\_connections = 3000    #wait\_timeout = 5022397   interactive\_timeout = 5022397    max\_connect\_errors = 1000   relay-log-recovery=1    #max\_allowed\_packet = 32M   sort\_buffer\_size = 4M    read\_buffer\_size = 4M   join\_buffer\_size = 8M    thread\_cache\_size = 64   #tmp\_table\_size = 256M    log\_slave\_updates=1   long\_query\_time = 1    slow\_query\_log = 1   skip-name-resolve    innodb\_buffer\_pool\_size=700M   #innodb\_data\_file\_path = ibdata1:1024M:autoextend   innodb\_flush\_log\_at\_trx\_commit=1    innodb\_log\_buffer\_size = 16M    innodb\_log\_file\_size = 256M   innodb\_log\_files\_in\_group = 2    innodb\_max\_dirty\_pages\_pct = 50    sync\_binlog=1    master\_info\_repository=TABLE   relay\_log\_info\_repository=TABLE    log\_timestamps=SYSTEM   gtid\_mode = ON    enforce\_gtid\_consistency = ON   master\_info\_repository = TABLE    relay\_log\_info\_repository = TABLE    log\_slave\_updates = ON    binlog\_checksum = NONE    log\_slave\_updates = ON   slave\_parallel\_type=LOGICAL\_CLOCK   slave\_parallel\_workers=8    slave-preserve-commit-order=on   #group\_replication\_compression\_threshold=200000   transaction\_write\_set\_extraction = XXHASH64   loose-group\_replication\_group\_name="01e5fb97-be64-41f7-bafd-3afc7a6ab555"   loose-group\_replication\_start\_on\_boot=off   loose-group\_replication\_local\_address="k8s3.mgrtest.mysqldb.svc.cluster.local:13306"   loose-group\_replication\_group\_seeds="k8s2.mgrtest.mysqldb.svc.cluster.local:13306,k8s3.mgrtest.mysqldb.svc.cluster.local:13306,k8s4.mgrtest.mysqldb.svc.cluster.local:13306"   loose-group\_replication\_bootstrap\_group = off   loose-group\_replication\_ip\_whitelist=10.96.0.0/16,172.16.0.0/16,192.168.0.0/16,172.17.0.0/16,127.0.0.1/8   report\_host = k8s3.mgrtest.mysqldb.svc.cluster.local   [mysqldump]    quick    max\_allowed\_packet = 32M | **MySQL参数配置文件configmap(节点3)** | | | --- | | cat mysql-mgr-cnf-2.yaml   apiVersion: v1 kind: ConfigMap metadata:  name: mysql-mgr-2-cnf namespace: mysqldb data:  mysql-mgr-2.cnf: |   [mysqld]    port = 3306    character\_set\_server = utf8    server\_id = 3    log\_bin = mysql-bin   relay-log = relay-bin    #back\_log = 500   #max\_connections = 3000    #wait\_timeout = 5022397   interactive\_timeout = 5022397    max\_connect\_errors = 1000   relay-log-recovery=1    #max\_allowed\_packet = 32M   sort\_buffer\_size = 4M    read\_buffer\_size = 4M   join\_buffer\_size = 8M    thread\_cache\_size = 64   #tmp\_table\_size = 256M    log\_slave\_updates=1   long\_query\_time = 1    slow\_query\_log = 1   skip-name-resolve    innodb\_buffer\_pool\_size=700M   #innodb\_data\_file\_path = ibdata1:1024M:autoextend   innodb\_flush\_log\_at\_trx\_commit=1    innodb\_log\_buffer\_size = 16M    innodb\_log\_file\_size = 256M   innodb\_log\_files\_in\_group = 2    innodb\_max\_dirty\_pages\_pct = 50    sync\_binlog=1    master\_info\_repository=TABLE   relay\_log\_info\_repository=TABLE    log\_timestamps=SYSTEM   gtid\_mode = ON    enforce\_gtid\_consistency = ON   master\_info\_repository = TABLE    relay\_log\_info\_repository = TABLE    log\_slave\_updates = ON    binlog\_checksum = NONE    log\_slave\_updates = ON   slave\_parallel\_type=LOGICAL\_CLOCK   slave\_parallel\_workers=8    slave-preserve-commit-order=on   #group\_replication\_compression\_threshold=200000   transaction\_write\_set\_extraction = XXHASH64   loose-group\_replication\_group\_name="01e5fb97-be64-41f7-bafd-3afc7a6ab555"   loose-group\_replication\_start\_on\_boot=off   loose-group\_replication\_local\_address="k8s4.mgrtest.mysqldb.svc.cluster.local:13306"   loose-group\_replication\_group\_seeds="k8s2.mgrtest.mysqldb.svc.cluster.local:13306,k8s3.mgrtest.mysqldb.svc.cluster.local:13306,k8s4.mgrtest.mysqldb.svc.cluster.local:13306"   loose-group\_replication\_bootstrap\_group = off   loose-group\_replication\_ip\_whitelist=10.96.0.0/16,172.16.0.0/16,192.168.0.0/16,172.17.0.0/16,127.0.0.1/8   report\_host = k8s4.mgrtest.mysqldb.svc.cluster.local   [mysqldump]    quick    max\_allowed\_packet = 32M | **准备pod文件节点1** | | | --- | | cat node1\_pod.yaml apiVersion: v1 kind: Pod metadata: name: k8s2  namespace: mysqldb  labels:   name: mysql-mgr spec:  nodeName: k8s2  hostname: k8s2  subdomain: mgrtest  containers:  - image: mysql:latest    name: mysql-mgr-0    imagePullPolicy: IfNotPresent        command: [ "/bin/bash", "-ce", "cd /usr/bin && mysqld --initialize-insecure --user=mysql && mysqld\_safe && tail -f /dev/null" ]    ports:      - containerPort: 3306   volumeMounts:    - name: tz-config      mountPath: /etc/localtime    - name: mysql-data      mountPath: /data/mysql/data/    - name: mysql-config     mountPath: /etc/my.cnf      subPath: my.cnf  volumes:   - name: tz-config      hostPath:        path: /etc/localtime    - name: mysql-data      hostPath:       path: /data/mysql/data/    - name: mysql-config     configMap:        name: mysql-mgr-0-cnf       items:          - key: mysql-mgr-0.cnf           path: my.cnf | **准备pod文件节点2** | | | --- | | cat node2\_pod.yaml   apiVersion: v1 kind: Pod metadata: name: k8s3  namespace: mysqldb  labels:   name: mysql-mgr spec:  nodeName: k8s3  hostname: k8s3  subdomain: mgrtest  containers:  - image: mysql:latest    name: mysql-mgr-1    imagePullPolicy: IfNotPresent    command: [ "/bin/bash", "-ce", "cd /usr/bin && mysqld --initialize-insecure --user=mysql && mysqld\_safe && tail -f /dev/null" ]    ports:      - containerPort: 3306   volumeMounts:    - name: tz-config      mountPath: /etc/localtime    - name: mysql-data      mountPath: /data/mysql/data    - name: mysql-config     mountPath: /etc/my.cnf      subPath: my.cnf  volumes:   - name: tz-config      hostPath:        path: /etc/localtime    - name: mysql-data      hostPath:       path: /data/mysql/data/    - name: mysql-config     configMap:        name: mysql-mgr-1-cnf       items:          - key: mysql-mgr-1.cnf           path: my.cnf | **准备pod文件节点3** | | | --- | | cat node3\_pod.yaml   apiVersion: v1 kind: Pod metadata: name: k8s4  namespace: mysqldb  labels:   name: mysql-mgr spec:  nodeName: k8s4  hostname: k8s4  subdomain: mgrtest  containers:  - image: mysql:latest    name: mysql-mgr-2    imagePullPolicy: IfNotPresent    command: [ "/bin/bash", "-ce", "cd /usr/bin && mysqld --initialize-insecure --user=mysql && mysqld\_safe && tail -f /dev/null" ]    ports:      - containerPort: 3306   volumeMounts:    - name: tz-config      mountPath: /etc/localtime    - name: mysql-data      mountPath: /data/mysql/data    - name: mysql-config     mountPath: /etc/my.cnf      subPath: my.cnf  volumes:   - name: tz-config      hostPath:        path: /etc/localtime    - name: mysql-data      hostPath:       path: /data/mysql/data/    - name: mysql-config     configMap:        name: mysql-mgr-2-cnf       items:          - key: mysql-mgr-2.cnf           path: my.cnf | ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_011.png) 3、创建namespace，service，configmap，pod | | | --- | | kubectl create -f namespace.yaml kubectl create -f service.yaml kubectl create -f mysql-mgr-cnf-0.yaml kubectl create -f mysql-mgr-cnf-1.yaml kubectl create -f mysql-mgr-cnf-2.yaml kubectl create -f node1\_pod.yaml kubectl create -f node2\_pod.yaml kubectl create -f node3\_pod.yaml | ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_011.png) 4、创建完成后查看状态 ### | | | --- | | 1.查看pods状态(看到node节点的mysql已经在运行) # kubectl get pods -n mysqldb -o wide NAME  READY   STATUS    RESTARTS   AGE    IP               NODE   NOMINATED NODE   READINESS GATES k8s2  1/1     Running   0          112s   172.16.109.65    k8s2             k8s3  1/1     Running   0          112s   172.16.219.1     k8s3             k8s4  1/1     Running   0          111s   172.16.106.129   k8s4 | ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_011.png) 5、配置MySQL8MGR **#节点1操作** | | | --- | | kubectl exec -it k8s2 -n mysqldb /bin/bash alter user root@localhost identified by Mysql123!@#; set sql\_log\_bin=0; create user rpl\_user@% identified by Rpl\_pass@123; grant replication slave on \*.\* to rpl\_user@%; flush privileges; set sql\_log\_bin=1; change master to master\_user=rpl\_user,master\_password=Rpl\_pass@123 for channel group\_replication\_recovery; install PLUGIN group\_replication SONAME group\_replication.so; reset master; set global group\_replication\_single\_primary\_mode=FALSE; set global group\_replication\_enforce\_update\_everywhere\_checks=TRUE; set global group\_replication\_bootstrap\_group=ON; start group\_replication; set global group\_replication\_bootstrap\_group=OFF; select \* from performance\_schema.replication\_group\_members; 1节点执行完后可以看到状态图 ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_007.png) | **节点2操作** | | | --- | | kubectl exec -it k8s3 -n mysqldb /bin/bash set sql\_log\_bin=0; create user rpl\_user@% identified by Rpl\_pass@123; grant replication slave on \*.\* to rpl\_user@%; flush privileges; set sql\_log\_bin=1; change master to master\_user=rpl\_user,master\_password=Rpl\_pass@123 for channel group\_replication\_recovery; install PLUGIN group\_replication SONAME group\_replication.so; reset master; set global group\_replication\_single\_primary\_mode=FALSE; set global group\_replication\_enforce\_update\_everywhere\_checks=TRUE; set global group\_replication\_recovery\_get\_public\_key=on; start group\_replication; select \* from performance\_schema.replication\_group\_members; ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_008.png) | **节点3操作** | | | --- | | kubectl exec -it k8s4 -n mysqldb /bin/bash set sql\_log\_bin=0; create user rpl\_user@% identified by Rpl\_pass@123; grant replication slave on \*.\* to rpl\_user@%; flush privileges; set sql\_log\_bin=1; change master to master\_user=rpl\_user,master\_password=Rpl\_pass@123 for channel group\_replication\_recovery; install PLUGIN group\_replication SONAME group\_replication.so; reset master; set global group\_replication\_single\_primary\_mode=FALSE; set global group\_replication\_enforce\_update\_everywhere\_checks=TRUE; set global group\_replication\_recovery\_get\_public\_key=on; start group\_replication; select \* from performance\_schema.replication\_group\_members; ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_009.png) | **配置完MGR后状态****为多主状态** ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_010.png) ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_011.png) 6、配置过程中问题及处理 ### **问题namespaces"mysqldb" already exists** 创建mysqldb的namespace时,因为已经存在一个mysqldb的namespace,报错如下: | | | --- | | kubectl create -f namespace.yaml Error from server (AlreadyExists): error when creating "namespace.yaml": object is being deleted: namespaces "mysqldb" already exists 查看状态也是Terminating的 ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_012.png) | **解决方式** | | | --- | | 1.获取mysqldb的namespace下json格式文件 # kubectl get namespace mysqldb -o json > /tmp/mysqldb.json 2.删除掉spec里面指定的内容 ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_013.png) 3.新开一个终端窗口执行kubectl proxy --port=8081,监听本地 # kubectl proxy --port=8081 Starting to serve on 127.0.0.1:8081 4.执行 # curl -k -H "Content-Type: application/json" -X PUT --data-binary @mysqldb.json http://127.0.0.1:8081/api/v1/namespaces/mysqldb/finalize 注意以上替换mysqldb名字 5.执行完后,查看namespace,mysqldb的Terminating状态已经没有了 ![](https://ucloud-blog.cn-bj.ufileos.com/articles/130047/images/130047_014.png) | ***End***